Table A1.

Qualitative review of additional studies and sleep variables not eligible for meta-analysis.

Study	T1D/controls (n)	Population	Study design	Sleep measurement	Sleep variables	Results
Barone et al. [35]	18/9	Adult	Matched case-control (age, BMI)	Sleep diary, actimeter, PSGd	Stages, duration, quality, OSA	No differences in percentage of sleep stages in control vs T1D participants: REM sleep (21.5% vs 19.9%, p = 0.62) or stage 3 (21.7% vs 20.6%, p = 0.76) Mean sleep duration from sleep diaries and nightly rest duration as assessed by actimeter were not correlated with mean glucose level as assessed by CGM.b However, in a subgroup of patients with low glycemia (mean glucose ≤154 mg/dL) (n = 9), nightly rest period negatively correlated with mean glycemia (r = −0.89, p = 0.03). Sleep quality as assessed by visual analogue scale did not correlate with mean glycemia as measured by CGM. In a subgroup of patients with low glycemia, sleep quality negatively correlated with mean glycemia. No differences in mean AHI (control vs T1D patients, 2.9 vs 3.4). None of the participants had OSA. In T1D, there was a correlation between AHI and mean glucose level (r = −0.55, p = 0.03), and arousal index and mean glucose level by CGM (r = 0.56, p = 0.03).
Blanz et al. [36]	93/93	Children/adolescents	Unmatched case-control	Interview as a part of psychiatric assessments (sleep disturbances)	Quality	More T1D reported sleep disturbances than control participants (χ2 test, 8.08, p < 0.01)
Borel et al. [16]	79/NA	Adults	Cross-sectional	Questionnaire evaluating OSA risk	OSA	Mean HbA1c was similar between those who reported snoring and those who did not snore (7.9% ± 1.0% vs 7.9% ± 1.1%, p = 0.89) (personal communication).
Caruso et al. [37]	49/36	Children/adolescents	Unmatched case-control	Questionnaire (Sleep Disturbance Scale for Children [SDSC])	Quality	T1D had significantly lower sleep quality than control participants (higher SDSC score). These included the total score (control vs T1D 43.8 vs 63.1, p < 0.001), disorders of initiating and maintaining sleep (55.0 vs 68.5, p < 0.001), disorders of sleep-wake transition (49.2 vs 57.1, p < 0.005) and disorders of excessive somnolence (48.5 vs 61.4, p < 0.001). No significant differences in the subscale of disorders of arousal, sleep hyperhidrosis, and sleep-disordered breathing between the two groups.
Donga et al. [38]	7/NA	Adults	Intervention study	Experimental sleep restriction	Sleep duration	Sleep restriction for one night (4 h) resulted in a significantly decreased glucose disposal rate during hyperinsulinemic euglycemic clamp (reflecting decreased insulin sensitivity) compared to a night with normal sleep duration (average 7.8 h).
Happe et al. [39]	46/50	Children/adolescents	Sibling study	Questionnaire	Quality, snoring, restless legs syndrome	No differences between T1D and control participants in percentages with restless legs syndrome symptoms (2.2% vs 2.0%), sleep initiation problem (10.9% vs 4.0%), sleep maintenance problem (6.5% vs 4.0%), or snoring (13.0% vs 14.0%)
Janovsky et al. [20]	20/22	Adults	Matched case-control (age, BMI)	PSGa	Stages, duration, OSA	No differences in percentage of sleep stages in control vs T1D participants: stage 1 (3.2% vs 4.5%), stage 2 (58.5% vs 57.8%), stage 3 (21.6% vs 21.2%) (personal communication). Sleep duration was similar between control participants vs T1D patients without CAN vs T1D without CAN (416 vs 379 vs 359 min) Mean AHI was similar between control participants vs those with T1D (3.7 vs 4.5). 40% of T1D vs 4.5% of control participants had OSA.T1D patients with CAN had significantly higher AHI than T1D patients without CAN (6.4 vs 3.2).
Jauch-Chara et al. [21]	14/14	Adults	Matched case-control (age, BMI, sex)	PSGc	Stages, duration	No differences in percentage of sleep stage 1 (controls vs T1D patients 19.2% vs 14.2%, p = 0.34), slow-wave sleep (controls vs T1D patients 14.9% vs 14.7%, p = 0.75). T1D patients spent more time in stage 2 than control participants (55.2% vs 47.2%, p = 0.01). During the first half of the night, there was a trend toward less time spent in slow-wave sleep in T1D patients than in controls (21.3% vs 24.7%, p = 0.09). Sleep duration was similar between the two groups (404 min vs 395 min, p = 0.93)
Kilmek et al. [30]	16,667/1,845,591	All ages	Cross-sectional, population based	Nationwide claims data on sleep disorders diagnosis (G47)	All sleep disorders in G47 diagnosis code	Sleep disorders were more commonly comorbid in T1D patients (relative risk = 1.9, 95% CI = 1.5–2.4).
Low et al. [11]	83/245	Adults	Matched case-control, comparable age and sex	Questionnaire (eight sleep questions from Autonomic Symptom Profile)	Quality	T1D patients had poorer sleep quality than controls (mean score = 0.27 vs 0.07; higher score reflects poorer sleep), but this was not statistically significant.
Matyka et al. [40]	15/15	Children/adolescents	Matched case-control (age, sex)	PSGa	Stages, quality	No significant differences in percentage of sleep stages between controls and T1D patients (stage 1: 4.3% vs 4.9%, p = 0.1; stage 2: 24.5% vs 22.7%, p = 0.7; stage 3: 51.4% vs 50.5%, p = 0.8) or REM sleep (17.8% vs 15.7%, p = 0.2). Compared to non-hypoglycemic night, the recordings during hypoglycemia revealed more time spent in slow wave sleep (60.4% vs 38.9%, p = 0.04) and less time in REM sleep (11.5% vs 15.2%, p = 0.04). T1D spent more time in short wake (<2 min) and long wake (>2 min) than controls (median 0.8% vs 0%, and 1.2% vs 0%, respectively)
Perfect et al. [23]	50/40	Children/adolescents	Matched case-control (age, BMI, sex)	PSGd (40 matched pairs) Actigraphyd (T1D cohort)	Stages, duration, quality, OSA	Compared to controls, T1D spent more time in stage 2 (57.2% vs 52.3%, p < 0.01) and less time in stage 3 (14.5% vs 18.9%, p < 0.05). More time spent in stage 2 was associated with higher HbA1c values and mean glucose levels by 5-day CGM. Sleep duration was not related to glucose control. Sleep duration each night was not related to waking glucose levels. No differences in sleep efficiency between T1D and control participants (86.6% vs 85.9%). Sleep efficiency was not related to glucose control. OSA prevalence and mean AHI were similar in T1D and control participants (35% vs 28%, and 2.48 vs 2.20), but central apnea index was higher in T1D patients than in controls (1.44 vs 0.33, p < 0.05). T1D patients with OSA (AHI ≥1.5) had significantly higher glucose levels by 5-day CGM than those without OSA. In addition, those with optimal glycemic control (n = 6) (HbA1c <7.5%) had lower AHI than those with suboptimal glycemic control (n = 34) (0.67 ± 0.49 vs 2.79 ± 4.64) (personal communication).
Perfect [22]	50/NA	Children/adolescent	Cross-sectional	Questionnaire (School Sleep Habit Survey)	Quality	Sleep quality was worse (as reflected by a lower score) in patients with suboptimal glycemic control (HbA1c ≥7.5%) than those with optimal glycemic control (7.8 ± 2.1, n = 42 vs 8.7 ± 1.3, n = 7). In addition, patients with suboptimal glycemic control had more daytime sleepiness (higher score) than those with optimal glycemic control (7.7 ± 3.5, n = 42 vs 5.2 ± 1.7, n = 7) (personal communication).
Pillar et al. [41]	15/15	Children/adolescents	Matched case-control (age, BMI)	PSGc	Stages, quality	No differences in percentage of sleep stage 3 (control vs T1D 25% vs 23%) or REM sleep (20% vs 20%). However, when analyzing only T1D with hypoglycemia during the recordings, T1D spent more time in stage 3 than controls (29% vs 25%, p < 0.05). No differences in sleep efficiency between the two groups. However, when analyzing only T1D with hypoglycemia, sleep efficiency increased compared to that in controls (95% vs 92%, p < 0.05)
Sivertsen et al. [12]	40/9843	Adults	Unmatched case-control	Questionnaire	Quality, OSA	No differences in sleep efficiency (as calculated from self-reported sleep timing and sleep latency) between control and T1D participants (85% vs 87%, p = 0.57) No differences in frequency of snoring and report of sleepiness at least three times/wk between control and T1D participants (4.1% vs 6.9%, p = 0.16)
Sturrock and Moriarty [31]	300/143	Adults	Unmatched case-control	Questionnaire (Nottingham Health Profile, sleep, NHP category)	Quality	T1D patients had worse sleep quality than control participants as reflected by a higher NHP sleep score (12.2 vs 9.3, p < 0.01)
van Dijk et al. [17]	99/99	Adults	Matched case-control (age, BMI, sex)	Questionnaires (OSA risk)	OSA	More T1D patients were at high risk for OSA compared to controls (17.2% vs 5.1%, p = 0.01). No association between OSA risk and suboptimal glycemic control (HbA1c ≥7.5%), OR = 0.50 (0.15–1.59), p = 0.24.
Varni et al. [33]	83/157	Children/adolescents	Unmatched case-control	Questionnaire (PedsQL Multidimensional Fatigue Scale)	Sleep quality	T1D had significantly worse sleep/rest fatigue score (as reflected by lower score) than control participants (69.3 vs 77.4, p < 0.05).
Villa et al. [42]	25/20	Children/adolescents	Matched case-control (age)	PSGa	OSA	Apnea index was higher in T1D than control participants (2.62 vs 1.40, p = 0.006). Central apnea index was higher in T1D with HbA1c ≥8% than control participants (2.54 vs 0.78, p < 0.0001), and tended to be higher in T1D patients with HbA1c <8% than in control participants (1.34 vs 0.78, p = 0.07). T1D patients with optimal glycemic control (n = 12) (HbA1c ≤7.9%) had a nonsignificant lower apnea index than those with suboptimal glycemic control (n = 11) (2.03 ± 1.78 vs 3.28 ± 1.64) and a significantly lower central apnea index (1.34 ± 1.29 vs 2.54 vs 1.27, p = 0.03).
Yeshayahu and Mahmud [32]	75/54	Children/adolescent	Unmatched case-control	Questionnaire	Duration	Mean sleep duration during weekdays was longer in T1D than in control participants (8.4 vs 8.0 h, p = 0.01), and both groups had longer sleep durations on weekends (an increase by 1.8 and 2.2 h in T1D and control participants, respectively). Mean sleep times or wake times in T1D patients did not differ based on HbA1c levels.

Abbreviations: AHI, apnea–hypopnea index; BMI, body mass index; CAN, cardiac autonomic neuropathy; CGM, continuous glucose monitor; OR, odds ratio; OSA, obstructive sleep apnea; PSG, polysomnography; REM, rapid eye movement; T1D, type 1 diabetes.

^aRecordings performed without glucose measurements.

^bRecordings performed with continuous glucose monitor.

^cRecordings performed under non-hypoglycemic conditions.

^dRecordings performed with glucose measurements. Some participants had hypoglycemia.